Photon Price Action Scanner [JOAT]Photon Price Action Scanner - Multi-Pattern Recognition with Adaptive Filtering
Introduction and Purpose
Photon Price Action Scanner is an open-source overlay indicator that automates the detection of 15+ candlestick patterns while filtering them through multiple confirmation layers. The core problem this indicator solves is pattern noise: raw candlestick pattern detection produces too many signals, most of which fail because they lack context. This indicator addresses that by combining pattern recognition with trend alignment, volume-weighted strength scoring, velocity confirmation, and an adaptive neural bias filter.
The combination of these components is not arbitrary. Each filter addresses a specific weakness in standalone pattern detection:
Trend alignment ensures patterns appear in favorable market structure
Volume-weighted strength filters out weak patterns with low conviction
Velocity confirmation identifies momentum behind the pattern
Neural bias filter adapts to recent price behavior to avoid counter-trend signals
What Makes This Indicator Original
While candlestick pattern scanners exist, this indicator's originality comes from:
1. Multi-Layer Filtering System - Patterns must pass through trend, strength, velocity, and neural bias filters before generating signals. This dramatically reduces false positives compared to simple pattern detection.
2. Adaptive Neural Bias Filter - A custom momentum-adjusted EMA that learns from recent price action using a configurable learning rate. This is not a standard moving average but an adaptive filter that accelerates during trends and smooths during consolidation.
3. Pattern Strength Scoring - Each pattern receives a strength score based on volume ratio and body size, allowing traders to focus on high-conviction setups rather than every pattern occurrence.
4. Smart Cooldown System - Prevents signal overlap by enforcing minimum bar spacing between pattern labels, keeping charts clean even when "Show All Patterns" is enabled.
How the Components Work Together
Step 1: Pattern Detection
The indicator scans for 15 candlestick patterns using precise mathematical definitions:
// Example: Bullish Engulfing requires the current bullish candle to completely
// engulf the previous bearish candle with a larger body
isBullishEngulfing() =>
bool pattern = close < open and close > open and
open <= close and close >= open and
close - open > open - close
pattern
// Example: Three White Soldiers requires three consecutive bullish candles
// with each opening within the previous body and closing higher
isThreeWhiteSoldiers() =>
bool pattern = close > open and close > open and close > open and
close < close and close < close and
open > open and open < close and
open > open and open < close
pattern
Step 2: Strength Calculation
Each detected pattern receives a strength score combining volume and body size:
float volRatio = avgVolume > 0 ? volume / avgVolume : 1.0
float bodySize = math.abs(close - open) / close
float baseStrength = (volRatio + bodySize * 100) / 2
This ensures patterns with above-average volume and large bodies score higher than weak patterns on low volume.
Step 3: Trend Alignment
Patterns are checked against the trend direction using an EMA:
float trendEMA = ta.ema(close, i_trendPeriod)
int trendDir = close > trendEMA ? 1 : close < trendEMA ? -1 : 0
Bullish patterns in uptrends and bearish patterns in downtrends receive priority.
Step 4: Neural Bias Filter
The adaptive filter uses a momentum-adjusted EMA that responds to price changes:
neuralEMA(series float src, simple int period, simple float lr) =>
var float neuralValue = na
var float momentum = 0.0
if na(neuralValue)
neuralValue := src
float error = src - neuralValue
float adjustment = error * lr
momentum := momentum * 0.9 + adjustment * 0.1
neuralValue := neuralValue + adjustment + momentum
neuralValue
The learning rate (lr) controls how quickly the filter adapts. Higher values make it more responsive; lower values make it smoother.
Step 5: Velocity Confirmation
Price velocity (rate of change) must exceed the average velocity for strong signals:
float velocity = ta.roc(close, i_trendPeriod)
float avgVelocity = ta.sma(velocity, i_trendPeriod)
bool velocityBull = velocity > avgVelocity * 1.5
Step 6: Signal Classification
Signals are classified based on how many filters they pass:
Strong Pattern : Pattern + strength threshold + trend alignment + neural bias + velocity
Ultra Pattern : Strong pattern + gap in same direction + velocity confirmation
Watch Pattern : Pattern detected but not all filters passed
Detected Patterns
Classic Reversal Patterns:
Bullish/Bearish Engulfing - Complete body engulfment with larger body
Hammer - Long lower wick (2x body), small upper wick, bullish context
Shooting Star - Long upper wick (2x body), small lower wick, bearish context
Morning Star - Three-bar bullish reversal with small middle body
Evening Star - Three-bar bearish reversal with small middle body
Piercing Line - Bullish candle closing above midpoint of previous bearish candle
Dark Cloud Cover - Bearish candle closing below midpoint of previous bullish candle
Bullish/Bearish Harami - Small body contained within previous larger body
Doji - Body less than 10% of total range (indecision)
Advanced Patterns (Optional):
Three White Soldiers - Three consecutive bullish candles with rising closes
Three Black Crows - Three consecutive bearish candles with falling closes
Tweezer Top - Equal highs with reversal candle structure
Tweezer Bottom - Equal lows with reversal candle structure
Island Reversal - Gap isolation creating reversal structure
Dashboard Information
The dashboard displays real-time analysis:
Pattern - Current detected pattern name or "SCANNING..."
Bull/Bear Strength - Volume-weighted strength scores
Trend - UPTREND, DOWNTREND, or SIDEWAYS based on EMA
RSI - 14-period RSI for momentum context
Momentum - 10-period momentum reading
Volatility - ATR as percentage of price
Neural Bias - BULLISH, BEARISH, or NEUTRAL from adaptive filter
Action - ULTRA BUY/SELL, BUY/SELL, WATCH BUY/SELL, or WAIT
Visual Elements
Pattern Labels - Abbreviated codes (BE=Engulfing, H=Hammer, MS=Morning Star, etc.)
Neural Bias Line - Adaptive trend line showing filter direction
Gap Boxes - Cyan boxes highlighting price gaps
Action Zones - Dashed boxes around strong pattern areas
Velocity Markers - Small circles when velocity confirms direction
Ultra Signals - Large labels for highest conviction setups
How to Use This Indicator
For Reversal Trading:
1. Wait for a pattern to appear at a key support/resistance level
2. Check that the Action shows "BUY" or "SELL" (not just "WATCH")
3. Confirm the Neural Bias aligns with your trade direction
4. Use the strength score to gauge conviction (higher is better)
For Trend Continuation:
1. Identify the trend using the Trend row in the dashboard
2. Look for patterns that align with the trend (bullish patterns in uptrends)
3. Ultra signals indicate the strongest continuation setups
For Filtering Noise:
1. Keep "Show All Patterns" disabled to see only filtered signals
2. Increase "Pattern Strength Filter" to see fewer, higher-quality patterns
3. Enable "Velocity Confirmation" to require momentum behind patterns
Input Parameters
Scan Sensitivity (1.0) - Overall detection sensitivity multiplier
Pattern Strength Filter (3) - Minimum strength score for strong signals
Trend Period (20) - EMA period for trend determination
Show All Patterns (false) - Display all patterns regardless of filters
Advanced Patterns (true) - Enable soldiers/crows/tweezer detection
Gap Analysis (true) - Enable gap detection and boxes
Velocity Confirmation (true) - Require velocity for strong signals
Neural Bias Filter (true) - Enable adaptive trend filter
Neural Period (50) - Lookback for neural bias calculation
Neural Learning Rate (0.12) - Adaptation speed (0.01-0.5)
Timeframe Recommendations
1H-4H: Best balance of signal frequency and reliability
Daily: Fewer but more significant patterns
15m-30m: More signals, requires tighter filtering (increase strength threshold)
Limitations
Pattern detection is mechanical and does not consider fundamental context
Neural bias filter may lag during rapid trend reversals
Gap detection requires clean price data without after-hours gaps
Strength scoring favors high-volume patterns, which may miss valid low-volume setups
- Made with passion by officialjackofalltrades
Pine Script® göstergesi
